1
Particle export during the Southern Ocean Iron Experiment (SOFeX)
Buesseler, K. O., J. E. Andrews, S. Pike, M. A. Charette
Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution
Woods Hole, MA 02543 USA
L. E. Goldson
University of East Anglia, Norwich, Norfolk, NR4 7TJ, UK
Now at: Plymouth Marine Laboratory, Plymouth, PL1 3DH, UK
M. A. Brzezinski
Department of Ecology Evolution and Marine Biology and the Marine Science Institute
University of California, Santa Barbara, CA 93106 USA
V. P. Lance
Duke University, Nicholas School of the Environment and Earth Sciences, Beaufort, NC 28516 USA
Revised for L&O ref#04-053
26 July 2004
2
ACKNOWLEDGEMENTS
SOFeX involved many collaborators without whom this study would not have been possible.
In particular, we thank K. Coale, K. Johnson, and the scientists and crews aboard the RV Revelle, RV
Melville, and USCG RIB Polar Star for their help and assistance. We could also not have written this
paper without the open sharing of SOFeX data produced by other investigators that are discussed here
or used in the figures; in this regard we particularly thank: E. Abraham, R. Barber, R. Wanninkhof, and
F. Chavez. This article benefited from comments by R. Anderson and another anonymous reviewer.
This work was supported by the Division of Ocean Sciences at the U.S. National Science Foundation,
the Biological and Environmental Research Office, Office of Science, U.S. Department of Energy, and
for RIB Polar Star logistical support, Polar Programs at the U.S. National Science Foundation.
3
ABSTRACT
We studied the effect of iron addition on particle export in the Southern Ocean by measuring
changes in the distribution of thorium-234 during a 4 week Fe enrichment experiment conducted in the
high-silicate high-nitrate waters just south of the Southern Antarctic Circumpolar Current Front at
172.5°W. Decreases in 234Th activity with time in the fertilized mixed layer (0-50m) exceeded those in
unfertilized waters, indicating enhanced export of 234Th on sinking particles after Fe enrichment. The
addition of Fe also affected export below the fertilized patch by increasing the efficiency of particle
export through the 100 m depth horizon. Extensive temporal and vertical Lagrangian sampling
allowed us to make a detailed examination of the 234Th flux model, which was used to quantify the
fluxes of particulate organic carbon (POC) and biogenic silica (bSiO2). Iron addition increased the
flux of both POC and bSiO2 out of the mixed layer by about 300%. The flux at 100 m increased by
more than 700% and 600% for POC and bSiO2, respectively. The absolute magnitude of the POC and
bSiO2 fluxes were not large relative to natural blooms at these latitudes, or to those found in
association with the termination of blooms in other ocean regions. Our results support the hypothesis
that Fe addition leads directly to significant particle export and sequestration of C in the deep ocean.
This is a key link between ocean Fe inputs and past changes in atmospheric CO2 and climate.
4
INTRODUCTION
The “iron hypothesis” (Martin 1990), suggests that inputs of the limiting micronutrient iron in
high-nutrient low-chlorophyll (HNLC) regions results in enhanced growth of marine phytoplankton.
This hypothesis has now been validated in many HNLC regions, including the Southern Ocean (Boyd
et al. 1999; Coale et al. 1996). However, it has been difficult to quantify whether or not this growth
response results in enhanced export of sinking particulate organic carbon to the deep ocean. Increased
new production and carbon export is the primary mechanism by which either natural or anthropogenic
iron enrichments could sequester carbon in the deep ocean and thus impact climate. The Southern
Ocean is of particular interest in this regard, as productivity in this region has been shown to be iron
limited (Boyd et al. 1999; Coale et al. 2004; Smetacek 2001), and surface waters return to depth with
high levels of macronutrients that could be utilized by phytoplankton if Fe supply were higher.
The Southern Ocean Iron Experiment (SOFeX) was a deliberate iron enrichment study that
involved 3 research vessels and a large group of scientists studying the biogeochemical response of the
surface ocean to iron at two sites; one north of the Polar Frontal zone in the low silicate waters of the
Subantarctic (Si(OH)4 <3 µmol L-1) and one farther south in the permanently high silicate waters to the
south of the Southern Antarctic Circumpolar Current Front, SACCF (Si(OH)4 ≈60 µmol L-1; Coale et
al. 2004). Prior studies in the Southern Ocean resulted in a diatom growth response after iron addition
to moderately high silicate waters (Boyd et al. 1999). A particularly significant finding of SOFeX was
the enhanced growth of a mixed community of larger cells (>5 µm diameter), including diatoms of the
genus Pseudonitzschia, dinoflagellates and the prymnesiophyte Phaeocystis, in the low silicate “North
Patch”. The growth response differed in the high silicate “South Patch” where up to 65% of the
phytoplankton in the bloom were diatoms (M. Landry, personal communication). A complete
5
overview of both experiments and the primary growth and geochemical responses can be found in
Coale et al. (2004).
Here we use the naturally occurring radionuclide thorium-234 (t ½ = 24.1 days) as an in-situ
tracer of particle flux during SOFeX. In this manuscript we focus on results from the South Patch and
examine the details of the changes in particle export as a response to iron. North Patch 234Th data were
too limited (n=2 occupations) to make any conclusions regarding changes in particle export. Newly
developed improvements in 234Th methods (Buesseler et al. 2001a) and a coordinated effort between
two ships over a 4 week period resulted in one of the most extensive sets of 234Th data ever collected.
A brief overview of these results has been presented previously (Buesseler et al., 2004), but herein we
expand upon these results with a detailed analyses of the time series evolution of the 234Th response in
context of larger biogeochemical data sets, and in terms of the 234Th flux model, its uncertainties, and
its application to the quantification of the fluxes of particulate organic carbon (POC) and other major
bioelements (biogenic silica, bSiO2 and particulate organic nitrogen, PON). We also compare these
data to prior observations of natural blooms at the SOFeX site and examine the export fluxes of C and
Si relative to uptake rates. Finally, the particle flux response in prior Southern Ocean fertilization
studies will be discussed.
SAMPLING AND ANALYSES
Study site and sampling protocols
Sampling took place over a 4 week period during occupation of the SOFeX South Patch in the
vicinity of 66oS, 172.5oW. Four separate iron enrichments of a single patch of water were made over a
grid that was initially 15 x 15 km, using methods described previously (Coale et al. 1998). In total,
1.26 x 103 kg of ferrous iron (as FeSO4) were added to surface waters (t0= 0730h GMT, 24 January
6
2002), resulting in Fe concentrations within the mixed layer that were expected to stimulate
phytoplankton growth (approx. 0.7 nmol L-1 Fe after addition). During the initial release, the inert
tracer SF6 was concomitantly added with the iron to mark the fertilized patch and to provide a
Lagrangian framework for the experiment (Watson et al. 1991). We use the convention that time zero
was the starting time of the initial Fe and SF6 release, though these additions took >24 hours to
complete.
Thorium-234 samples were collected between 30 January and 14 February 2002 on the RV
Melville (t= day 5.8 – 21.6) followed by the USCG RIB Polar Star between February 15 – 20 (t= day
21.7 – 27.4). During each cruise, samples were collected both inside the iron fertilized patch (hereafter
referred to as “IN”), and at control stations outside the patch (hereafter referred to as “OUT”). The
presence of high SF6 at a given station was used to differentiate between IN vs. OUT stations.
Profiles of total 234Th were collected using 4 liter samples taken from Go-Flow® or Niskin
bottles deployed by rosette or as single bottles strung on Kevlar line within the upper 100-150 m on 18
separate casts. On the RV Melville, 234Th sampling was conducted primarily using a trace metal clean,
“TM” rosette used for primary production and other biological rate measurements and studies. SF6
was not measured on every TM rosette cast, though SF6 was used to identify IN vs. OUT stations on at
least selected surface samples at the same station. On the RIB Polar Star, SF6 was measured on every
cast where 234Th samples were collected.
Samples for particulate 234Th, POC and bSiO2 were collected using paired battery powered
large-volume in situ pumps (McLane Labs, Falmouth, MA, USA) deployed to predetermined depths,
usually one within the mixed layer (typically ~20 m) and one below (~50-120 m; n=21 pump casts).
Sample volumes on the order of 400 liters (measured at the outlet via an in line mechanical flow meter)
were passed sequentially through 142-mm diameter filters mounted in a PVC baffled filter housing at
7
flow rates of 8 L min-1. The first filter was a polyester screen with a nominal mesh size of 54 µm.
This was followed by a quartz filter (QMA) with a 1 µm nominal pore size.
Shallow samples were also collected along several transects from depths of either 25 m or the
surface. Surface samples for 234Th, SF6 and chlorophyll a (Chl a) were collected from the ship’s
uncontaminated surface seawater line or a bucket. At selected transect stations 25 m samples were
collected using either a Niskin bottle or TM rosette.
234Th
analyses
Within 1 hour of collection, unfiltered 4 liter samples for total 234Th determination were
acidified to a pH  2 with 6 ml of concentrated HNO3. After shaking the samples vigorously, 0.3 ng of
thorium-230 was as a yield monitor. The sample was shaken again and allowed to equilibrate for 12
hours. After equilibration, the pH was adjusted to 8 ± 0.15 with concentrated NH4OH. Extraction of
Th from the samples was accomplished by coprecipitation of Th via formation of MnO2 precipitate
formed by the addition of KMnO4 and MnCl2 solutions at pH = 8 (Buesseler et al. 2001a; BenitezNelson et al. 2001). Samples were vigorously shaken after the addition of each reagent and after the
formation of the MnO2 precipitate, allowed to stand for 12 hours followed by filtration on to 25-mm
diameter microquartz filters. Finally, the filtered Mn precipitates were dried overnight and mounted
under one layer of Mylar film and two layers of standard Al foil in preparation for beta counting.
Particles collected on the 54 µm pore size screens were rinsed immediately after collection onto
single 25 mm diameter Ag filters (Sterlitech Corp., 1.2 µm nominal pore size) using prefiltered deep
water. The Ag filters were dried over night at 60oC and prepared for beta counting with Mylar and Al
covers identical to the 234Th samples. The 142 mm diameter QMA filters were dried overnight and 25
8
mm diameter punches (n=20; equivalent to 52% of the active filter area) were cut, stacked and pressed
into inverted beta mounts and prepared for counting.
Analyses of particulate and total 234Th activities were performed at sea using gas-flow
proportional beta counters manufactured by Risø National Laboratories (Roskilde, Denmark)
following methods described in Buesseler et al. (2001a). Initial at sea beta counting for 6-24 hours
was followed by a final background count in the lab after decay of 234Th for at least 6 half-lives. Initial
count rates were typically 2-4 counts per minute (cpm) for total 234Th and final count rates averaged
0.5 cpm, relative to the detector backgrounds of 0.15 – 0.20 cpm. For total 234Th samples, anion
exchange chemistry was performed and recovery of our added 230Th yield monitor was quantified by
ICP-MS with addition of a 229Th internal standard (Pike et al. 2004). Corrections were applied to 234Th
activity calculations based on 230Th recoveries. All data are decay corrected to the time of collection
and reported with a propagated error that includes counting uncertainty and uncertainty associated with
sample volumes and other calibration errors. Errors were typically ± 0.04-0.06 dpm L-1 on total 234Th.
A recommended check on overall accuracy of the total 234Th activity measurement is through
analyses of deep waters where equilibrium with 238U is assured (i.e. 234Th = 238U, with 238U being
derived from salinity, Chen et al. 1986). On the RV Melville, we analyzed 8 samples from 1400 m and
found a 3.0% standard deviation in 234Th activity. On the RIB Polar Star, we were only able to collect
two 4L samples from 500m, and the 234Th activities were identical to 238U within our error of ± 2%.
Having made this calibration, one can be certain that the 234Th:238U ratio in the surface, accurately
reflects the 234Th:238U disequilibrium that is used to calculate the fluxes that are central to the
application of this tracer.
9
Particulate organic carbon, organic nitrogen and biogenic silica
All samples for POC, PON and bSiO2 analyses were taken from the same filter matrix that had
been analyzed for 234Th enabling the particulate ratio of each element to 234Th to be determined. Thus,
the >54 m particles, which had been rinsed on to 25mm diameter Ag filters and counted for 234Th,
were subsequently subsampled by weight for POC, PON and bSiO2 analyses. Subsamples of the QMA
filter were obtained by removing a small punch from the 142 mm diameter filter, and analyzing using
standard methods for POC and PON. Prior to CHN analyses, all samples were fumed with HCl to
remove carbonate (Pike and Moran 1997). CHN analyses were performed using a Thermo Quest Flash
EA 1112 (CE Instruments, Italy). Biogenic silica was not measured on the QMA fraction due to the
high Si blanks associated with this filter matrix (Buesseler et al. 2001a). Details of the NaOH
digestion methods and analyses used to determine bSiO2 on Ag filters are described in Brzezinski et al.
(2001) and Buesseler et al. (2001b).
Chlorophyll and primary production
For the SOFeX South patch, chlorophyll and primary productivity were measured from patch
days ~-1.5 to ~19.5 (from aboard the RV Roger Revelle: t=-1.5 to 9.5 and aboard the RV Melville:
t=~7 to ~19.5). Surface chlorophyll was measured aboard the USCG RIB Polar Star for patch days
~21.5 to 27.5. Selections of these chlorophyll and productivity data have been presented in Coale et al.
(2004), and methods are only summarized here. Chlorophyll a samples were collected from a standard
rosette (or shipboard underway system in the case of Polar Star) and filtered onto Whatman (GF/F)
glass fiber filters, extracted with 90% acetone for 24 hours in the dark at –20ºC (Venrick and Hayward
1984) then analyzed fluorometrically (Holm-Hansen et al. 1965; Lorenzen 1966). Primary
productivity samples were collected from a trace-metal clean rosette (Hunter et al. 1996) using clean
10
techniques to minimize trace-metal contamination and incubated on-deck (Barber et al. 1996). Water
samples were collected into 75 ml sterile polycarbonate tissue culture flasks, inoculated with 10µCi of
14
C-carbonate solution and incubated for 24 hours in incubators screened with blue plastic plus neutral
density screening to achieve a given percentage of surface incident irradiance (Ed(0-)), nominally 100,
47, 16, 10 and 1%. Incubator temperatures were maintained by a continuous flow of surface seawater.
Incubated samples were harvested on Whatman (GF/F) filters, acidified with 0.5ml of 0.5N HCl for 24
hours, scintillation fluid added; then, after another 24 hours in the dark at room temperature, sample
activities were counted in a liquid scintillation counter. Time-zero controls and total added activity
were determined similarly by variations of the above methods (Huntsman and Barber 1977).
RESULTS
Transect data
The main objective of this study was to determine if the Fe-induced increase in plankton
biomass also leads to an increase in particle export flux, as evidenced by decreased 234Th activities. As
a means of providing a rapid assessment of these impacts, we conducted a series of surface transects
across the Fe-fertilized patch. During each transect, the boundaries of the patch were clearly
identifiable by the SF6 data, which increased from background to maximum SF6 concentrations at the
patch center (background surface SF6 levels OUT are ~1-2 fmol L-1 due to global releases; Figure 1).
Note that the maximum SF6 concentration along each transect decreases over time from ~40 fmol L-1 to
5 fmol L-1. This is due to dilution from horizontal patch dispersal, mixed layer deepening and losses
across the air-sea interface (Goldson, 2004). Recall that SF6 was only added with the first of four iron
additions. While every effort was made to overlay each Fe addition within the region tagged with SF6
some enrichment of waters without SF6 likely occurred.
11
Chlorophyll a concentrations increased over time within the fertilized patch, providing first
order evidence that iron addition to these waters stimulated phytoplankton production, as discussed in
Coale et al. (2004). However, also note in the satellite image from day 23 (Figure 2; only clear
satellite image during SOFeX) and on the transect data from days 25 and 28 (Figures 1E and 1F) that
high chlorophyll concentrations were also observed in unfertilized waters surrounding the patch. For
example, on the day 25 transect (Figure 1E), there was a general trend towards slightly higher 234Th
activities in the northern end of the transect between 65.3 to 65.5°S, where Chl a concentrations were
~1.5 mg m-3, vs. a maximum of ~2 mg m-3 at patch center. Also, on the day 28 transect (Figure 1F),
high Chl a was observed along the western edge (1.6 mg m-3), decreasing outside the patch towards the
eastern most stations, where some of the highest 234Th activities were found.
The clear relationship between iron addition and Chl a in the surface transect data was not
reflected in 234Th, which displayed relatively invariant activities across the patch center along every
individual transect (Figure 1). However, there was a subtle decrease in average 234Th activities in the
patch relative to the 238U activity over time (to be discussed below in the context of the vertical
profiles), and some small scale variability in 234Th along the transect that is unrelated to the iron
fertilized patch.
The sampling of both high and low Chl a waters and more variable 234Th activities outside the
patch exemplifies the natural heterogeneity of the Southern Ocean during the austral summer. This
heterogeneity also highlights the difficulty in defining an OUT control station among these local
variations in biomass, 234Th and biogeochemistry. While we lump all profiles from low SF6 waters as
OUT stations, this variability should be kept in mind as we compare temporal changes in the OUT
control stations with the Lagrangian time series sampling of the SF6 tagged IN stations.
12
Time series profiles
At the start of SOFeX (t=5.8 d), there was a sharp pycnocline at the base of the mixed layer
(30-35 m), below which a broad temperature minimum extended from the base of the mixed layer to
80 m (Figure 3A). Density changes are driven primarily by salinity, which increased gradually below
the mixed layer. These T/S properties are common to the Southern Ocean waters south of the SACCF
along 170oW (Morrison et al. 2001) and were seen at all stations inside and outside the patch. The T/S
structure is formed after the local retreat of the sea ice in the spring (starting around late November at
this location), whereby warming of the lower salinity melt waters creates a gradually shoaling and
warmer mixed layer overlying the near freezing waters below. During SOFeX, the depth of the mixed
layer increased over time to ~50 m both IN and OUT of the patch due to a series of storm events
(Goldson 2004). These storms also reduced the vertical extent of the temperature minimum. The
hydrographic changes recorded in the two final IN and OUT profiles from SOFeX (Figures 3B and 3C)
are characteristic of the changes observed in the complete data set (Coale et al. 2004; Bishop et al.
2004).
At the start of SOFeX, nutrient concentrations were high (NO3>25 µmol L-1; Si(OH)4 >60
µmol L-1) and biomass was low (Chl a <0.5 mg m-3), typical of austral summer conditions for this
HNLC region (Morrison et al. 2001). Previous SeaWiFS satellite estimates of surface chlorophyll
along 170oW suggest that higher Chl a and primary productivity would have been found in December
and early January in association with ice edge blooms prior to the initiation of SOFeX (Buesseler et al.
2003). During these early blooms, maximum Chl a concentrations on the order of 0.8-1.4 mg m-3 have
been seen, which are roughly twice the values seen here at the start of SOFeX. The relatively low 234Th
activities observed in the OUT stations also suggest that the experiment began subsequent to previous
particle export likely associated with the December/January ice edge blooms.
13
One of the major characteristics of SOFeX and previous Southern Ocean iron fertilization
experiments has been a slow response to iron in terms of biomass accumulation, due mainly to the
slower metabolism of phytoplankton in the cold Antarctic waters. Thus while a physiological response
to Fe, as indicated by an increase in photosynthetic efficiency, was observed after only a few days,
increases in biomass were not detected until 5-7 days after Fe was first added (Boyd and Abraham
2001; Coale et al. 2004). During SOFeX, biomass increased to >2-3 mg m-3 Chl a within the patch
after 15-20 days but remained <0.5 mg m-3 Chl a outside the patch throughout the experiment.
At the start of SOFeX (t=5.8 d), the lowest 234Th activities were observed within the mixed
layer (Figure 3A). This initial 234Th:238U disequilibrium indicates a previous flux of particles
associated with the natural bloom cycle. Thorium-234 activities increased with depth to values greater
than 238U (>2.4 dpm L-1), indicative of remineralization. Thus at the start of SOFeX, the site had
already experienced a vertical flux of particles carrying 234Th out of the mixed layer, accompanied by
the remineralization of some of these particles at depths of 80-120 m.
Overall, upper ocean (0-50 m) 234Th activities ranged between 1.7 and 2.1 dpm L-1 during the
course of the experiment and large changes over time were not obvious. To put these 234Th data in
context, we have seen decreases in 234Th activities as large as 1 dpm L-1 associated with the end of the
spring bloom in the North Atlantic (Buesseler et al. 1992) and the end of the SW Monsoon in the
Arabian Sea (Buesseler et al. 1998). Thus the lack of a stronger inverse correlation between surface
SF6 and 234Th, is evidence that during the 4 week occupation of the SOFeX bloom, surface ocean
export was not particularly elevated, or at least small relative to other observations of particle export
associated with natural blooms from other regions.
The IN station time series was characterized by a small decrease in mixed-layer 234Th and a
penetration of the growing deficit to greater depths over the course of the experiment (see time series
14
integrals below). By the end of SOFeX, mixed layer 234Th activities had decreased slightly both IN
and OUT (Figures 3B and 3C). However, there were two important differences observed between the
IN and OUT station time series: (1) the vertical extent of the 234Th disequilibrium was greater at IN
stations, and (2) the excess 234Th feature at 80-120 m present at the beginning of the experiment
persisted at OUT stations, but eventually disappeared at IN stations.
To rule out physical mixing processes as a potential explanation for these changes, we plotted
the final IN and OUT 234Th activities vs. density (Figure 4). Within the lowest density surface waters,
there was little difference in 234Th activity, consistent with the lack of IN vs. OUT differences observed
in context of the surface transect data. By the end of SOFeX we observed 0.1-0.2 dpm L-1 lower total
234
Th activities (mean error = 0.05 dpm L-1) within the patch extending to densities of 27.7, which
includes depths below the mixed layer down to the peak in 234Th excess at 80-120m. These changes
cannot be explained by subtle differences in the IN vs. OUT hydrography, and must therefore be a
result of the SOFeX Fe release. We conclude that this small decrease in total 234Th activity over time
is direct evidence of enhanced particle flux within the SOFeX patch.
As noted above, an unusual feature of the early IN stations and all OUT stations was the
presence of excess 234Th at depths of 80-120m (Figures 3A and 3C). Since the only in-situ source of
234
Th in seawater is 238U decay, total 234Th activities can never exceed those of 238U unless there is
some additional non-local source, in this case from sinking particles either releasing 234Th to the
dissolved phase or disaggregating in to suspended material. Interestingly, excess 234Th remained
unchanged in the OUT stations, yet for the IN station the excess 234Th generally disappeared by the end
of the experiment (Figure 5). This indicates that some process below the mixed layer responded to the
development of the bloom, by allowing a greater fraction of the 234Th on sinking particles to reach
greater depths. This implies either a change in the nature of the sinking particles, i.e. faster settling,
15
less labile, and/or a subsurface change in the processes responsible for remineralization, for example a
migration of zooplankton which were consuming sinking particles, to shallower depths as the bloom
developed within the mixed layer.
The reproducibility of the excess thorium feature in the OUT stations and the temporal changes
IN, provide unique insights in to the extent of remineralization below both the natural and iron-induced
biological communities. The excess 234Th does not tell us the exact process responsible for the
decrease in particle flux, be it physical-chemical, microbial or zooplankton driven, but further
examination of this feature in context of the broader SOFeX data sets will be warranted, once these
results are available. However it should be noted that the SOFeX sampling program was focused
primarily on the expected surface water response to iron, and not this subsurface change in the particle
cycle, thus no specific experiments designed to follow this deeper feature were conducted.
From this detailed time series data set, we can quantify temporal changes in particle flux
derived from 234Th:238U disequilibrium. All of the SOFeX total 234Th profiles from the South Patch are
plotted in Figures 5A and 5B and the data provided in Web Appendix 1 at
http://www.aslo.org/lo/toc/vol_xx/issue_x/xxxxa1.pdf. This single data set comprises the most
extensive suite of 234Th data ever collected at a single site within a relatively short four week time
period. Since the IN data were collected as part of the SOFeX Lagrangian sampling of the fertilized
patch, it also provides us with a true time series record of the development of the SOFeX bloom.
Time series integrals
The flux of 234Th on sinking particles can be determined by solving for the balance of 234Th
sources and sinks, using the mathematical expression:
 234Th / t ( 238U  234Th)    P  V
(1)
16
where Th/t is the change in 234Th activity with time, 238U is the uranium activity determined from its
conservative relationship with salinity (238U (dpm L-1) = 0.071●salinity; Chen et al. 1986), 234Th is the
activity of total 234Th,  is the decay constant for 234Th (= 0.0288 day-1), P is the net export flux of
234
Th on sinking particles, and V is the sum of advective and diffusive 234Th fluxes.
From the integral, or depth averaged 234Th and 238U activities, P (in terms of dpm m-2 d-1) can
be solved for any given depth horizon. Typically with single profiles of 234Th at a given site, steady
state is assumed (Th/t = 0). As can be seen mathematically in Eq. 1, a negative Th/t requires a
larger 234Th flux in order balance the production and decay rates. The changes in 234Th activity
(Th/t) following oceanic blooms can be one of the largest terms in Eq. 1 and should always be
considered in these settings (Buesseler et al. 1992; Cochran et al. 1997; Tsunogai et al. 1986; Wei and
Murray 1992). Thus, a time series sampling approach is essential when applying 234Th in open ocean
iron enrichment studies, given the expectation of a particle flux response to the iron addition (Bidigare
et al. 1999; Charette and Buesseler 2000).
In order to calculate fluxes over time at different depths from a non-uniform vertical sampling
grid, the average total 234Th activities (dpm L-1) were binned into five depth integrals (0-25 m, 0-50 m,
0-75 m, 0-100 m, 0-125 m). We do this since to calculate the flux at 25m, one needs to know the
average 234Th activity from the surface to 25m, for 50m, from the surface to 50m, etc. As detailed in
Buesseler et al. (1992; Appendix A) to estimate Th/t for a non-steady state (NSS) solution to Eq. 1,
one can use the linear change in 234Th activity over the course of time series measurements to estimate
Th/t. For the IN data, we measured a statistically significant linear decrease in average total 234Th
activities vs. time during SOFeX for all depth integrals (Figure 6). As expected from the profiles, the
average 234Th activity increased with the deeper depth integrals. Also, the time series decreases were
greatest for the 0-75 and 0-100 m integrals IN, which include the interval where the 234Th
17
remineralization peak progressively disappeared (Table 1). While there is some variability around this
slope, especially in the upper 25-50 m, the decrease over 28 days is statistically greater than the
measurement error and most data fall within the 95% confidence limits of the linear fit. The calculated
standard error on the slope of these curves ranges from <10% for the 0-125 m depth integral to 35%
for the 0-25 m layer (Table 1).
The results in Table 1 allow us to quantify Th/t in Eq. 1 and the error associated with this fit,
so that we can apply a NSS solution to Eq. 1 for the IN stations. The OUT data are also plotted (Figure
6) but no linear fit is drawn as none of these data had slopes statistically different from zero (R2<0.15).
This lack of change in Th/t OUT is one of the biggest uncertainties in the interpretation of the OUT
flux data, since as seen in these transects, there is natural variability in 234Th outside of the patch, and a
true Lagrangian control was not sampled.
Thorium-234 fluxes
Using the time series integrals of total 234Th activity and Eq. 1, we can calculate the evolving
234
Th flux patterns with depth and time at both IN and OUT stations. For both IN and OUT stations,
we assume steady state conditions at the start of SOFeX (i.e. Th/t = 0) and use the measured 234Th
activity for any given depth integral to calculate the flux. For the IN stations, we applied a NSS
solution to Eq. 1 beginning with day 10 (we assume there was no significant decrease in 234Th until
this second sampling point). We use the linear fit to the 234Th activity decrease with time as our
estimate of Th/t during the entire experiment (Table 1), and to calculate the average 234Th activity
for any depth integral on a given day. For OUT stations there was no clear 234Th decrease with time,
hence we calculated the 234Th flux using a steady state model throughout the experiment. Errors on the
18
fluxes are derived from propagating errors from the slope Th/t and the error on the average 234Th
activity.
Using these assumptions the flux profiles at the start (day 6 IN and 7 OUT), day 10, day 20 and
at the end (day 27) are plotted in Figure 7A and 7B. At the start of SOFeX, IN and OUT fluxes were
similar, ranging from 250-300 dpm m-2 d-1 at 25 m, increasing to a maximum of 500-700 dpm m-2 d-1
at 75m and decreasing below to values of 300-400 dpm m-2 d-1 at depths of 100-125 m (Figure 7A and
7B). The largest increase in 234Th flux with depth is found between 25 and 75 m for the IN stations
(Figure 7A). Compared with the start of the experiment (at both IN and OUT stations), 234Th fluxes at
the end between 75-125 m increased nearly five fold to 2300-2800 dpm m-2 d-1. At the OUT stations,
the 234Th flux distribution with depth was similar over time and the magnitude of the 234Th flux
remained low (Figure 7B).
The most significant finding of this study was the observed 234Th flux increase IN, which is
strong evidence of enhanced particle export in response to an open ocean iron fertilization experiment
(Buesseler et al. 2004). In addition, there was an unexpected change in the 234Th flux below the ironenriched waters. While the 234Th remineralization peak at 80-120 m persisted at the OUT stations for
the entire experiment, this feature essentially disappeared at the IN stations by day 15.
There are several important implications for these findings. Not only did iron fertilization
enhance particle flux within the upper fertilized layers, but it also may have influenced the fate of these
materials once they leave the surface waters. In the latter case, elements that left the upper 50 m in
association with sinking particles may be more efficiently transported to depth after fertilization.
Possible explanations for this effect include changes in heterotrophic processes responsible for particle
break-up and consumption, and/or changes in the nature of the material leaving the euphotic zone that
may lead to more efficient transport to depth (i.e. more rapidly sinking, or less labile particles).
19
Impact of physical transport on 234Th flux assessment
The influence of vertical mixing on the 234Th flux model can be substantial, particularly in high
upwelling sites and times, such as the coastal Arabian Sea at the onset of the monsoon (Buesseler et al.
1998), or in equatorial regions, where upwelling can increase the calculated 234Th fluxes by up to a
factor of two (Buesseler et al. 1995; Dunne and Murray, 1999). Horizontal transport of 234Th has also
been shown to be particularly important in coastal regions (Benitez-Nelson et al. 2000; Charette et al.
2001; Gustafsson et al. 1998). With the exception of loss across the air-sea interface, SF6 behaves
conservatively in seawater. As such, the temporal evolution of the SF6 tracer along and across
isopycnals can be used to determine the horizontal and vertical exchange rates, respectively, which are
applied to 234Th.
During SOFeX, the temporal spread of SF6 across the lower boundary of the mixed layer was
used to derive a rate of vertical diffusivity of approximately 0.3 cm2 s-1. Details of this study are
presented in Goldson (2004). Since the activity of 234Th increases with depth, supply of 234Th via
vertical processes would provide a source of additional 234Th to shallower layers and thus by ignoring
this process we would underestimate vertical fluxes. Given this rate of vertical diffusion and the
vertical gradient of 234Th below the surface disequilibrium, we estimate that vertical mixing processes
could supply ~100 dpm m-2 d-1. At 100 m inside the patch at the end of SOFeX, this vertical source of
234
Th to the mixed layer was <5% of the total 234Th flux and fell within the overall error of the model.
Patch dilution can help us estimate the impact of horizontal mixing on the 234Th activity
balance. Dilution has been calculated for SOFeX from both physical strain methods (Abraham et al.,
2000) and by looking at the losses of SF6 over time, corrected for outgassing (Coale et al. 2004). Both
methods agree fairly well, and here we use an average dilution rate of 0.05 d-1 derived from SF6 (range
20
0.03-0.07 d-1). Dilution can affect geochemical budgets differently, depending upon the time-series
response and IN-OUT gradients, and in the case of 234Th, the relative magnitude of the physical
transport terms vs. the production and decay rates of 234Th (Eq. 1). Using the 0-50 m integrated total
234
Th data (Figure 6) to calculate weekly IN-OUT gradients, we estimate that dilution would impact
our 50m fluxes by –140, + 50 and +240 dpm m-2 d-1 during weeks 2, 3 and 4, respectively. The
magnitude of this adjustment is small relative to the NSS fluxes and since we do not have adequate 4D coverage to assess horizontal gradients very reliably, we choose not to adjust our NSS fluxes by
these small corrections. Note also that total errors on the elemental fluxes are ±25-50%, so we don’t
consider these dilution effects on the 234Th flux to be significant within these errors.
Flux of particulate organic carbon
If the flux of 234Th on sinking particles is known, then the export flux of POC and other
constituents can be estimated from knowledge of the ratio of a given constituent, E, to 234Th in
particles, using:
flux E = E/234Th • flux 234Th
(2)
This empirical approach assumes that we know the flux of 234Th accurately, and that the particulate
E/Th ratio is representative of sinking particles at the same time and depth of interest. During SOFeX,
the elemental ratios of POC, PON and bSiO2 vs. 234Th on sinking particles were derived from sizefractionated samples collected using in-situ pumps (Web Appendix 2 at
http://www.aslo.org/lo/toc/vol_xx/issue_x/xxxxa2.pdf). For POC/Th, we now have considerable
background on its variability (Buesseler 1998) and generally find decreasing ratios with depth. Higher
ratios are commonly found during blooms and at high latitudes, where large diatoms and Phaeocystis
blooms are common. While the exact processes that control this ratio are not known, these two basic
21
features can be explained by preferential recycling of POC from particles as they sink leading to lower
POC/Th, and higher POC/Th being associated with larger cells that have lower surface area to volume
ratios, since 234Th is a surface reactive element and POC varies as a function of cell volume. We
assume that the larger particle size classes have higher settling velocities and use this ratio in our
elemental flux calculations, but also examine the finer suspended material to put some bounds on
possible errors associated with this assumption.
During SOFeX, time series particulate samples were commonly collected in pairs, one sample
from within the mixed layer and one from below. At some of the later stations, we also had 3 or 4 point
depth resolution. POC/Th ratios averaged just under 10 µmol dpm-1 in the upper 50 m on the larger
>54 m size class, with little change during the course of SOFeX (Figure 8A). For the small particles
(>1-54 µm), we observed similar ratios to the large particles for the first 2-3 weeks of SOFeX, but
strikingly higher ratios > 30 µmol dpm-1 during the 4th week of the experiment. At depth, the same
ratios were much lower (4.4 µmol dpm-1) and similar between the two size classes throughout these
observations (Figure 8C). The OUT data displayed more similar POC/Th ratios between the two
particle size classes and with depth (Figure 8B and 8D). There was evidence of slightly higher
POC/Th late in the upper 50m OUT, but not as large as the changes IN.
We attribute the temporal increase in POC/Th ratios IN to changes in the iron-induced
phytoplankton bloom, since the community response was greatest for the larger, >20 m diatoms after
the addition of iron (Coale et al. 2004). If the material caught on the 1m filter represents fresh
biogenic material between 1-54 µm, then it implies that over time, the increase in POC/Th attributable
to these new diatom cells, had not yet impacted the larger aggregates in the >54 m fraction. Also,
there was no evidence of the POC/Th ratio increase at depth over time, suggesting that the diatoms had
not yet aggregated or sunk to depth. This result is consistent with the observation that the SOFeX
22
bloom had not reached a stage of high export characteristic of bloom termination. This also agrees
with the general findings that Chl a and POC concentrations had leveled off by week 3-4, and there
was a lack of any measurable Fe stress (as evidenced by continued high photosynthetic efficiency)
even at the end of our 28 day observation period (Coale et al. 2004).
As early as day 6, when there was not yet any evidence of a decline in total 234Th activity
within the patch, particulate 234Th activities >1-54 m had decreased slightly in the upper 50 m (0.35
dpm L-1 IN vs. 0.53 dpm L-1 OUT on 1-54 m particles; Web Appendix 2 at
http://www.aslo.org/lo/toc/vol_xx/issue_x/xxxxa2.pdf). During a previous Southern Ocean iron
fertilization experiment we observed a similar decrease in the partitioning of 234Th onto smaller
particles (Charette and Buesseler 2000) and can only speculate as to the cause. One response to added
Fe may be an increase in production of organic ligands by the biota that complex Fe and other trace
metals (Bowie et al. 2001). It is possible that these ligands were competing for adsorbed thorium
atoms, which would explain the decrease of 234Th in the particulate pool (Santschi et al. 2003). This
IN patch decrease in particulate 234Th activities coincided with a smaller IN patch increase in POC
concentration (2.5 to 3.0 µmol L-1 by day 6), therefore the higher shallow POC/Th ratios at IN stations
were driven by both of these concentration changes. A change in 234Th partitioning does not impact
the 234Th flux directly, and the total activities IN and OUT were similar during the first 1-2 weeks of
the experiment.
We assume from theory that the larger particles are the best representation of the sinking
particle pool, and thus use the >54 m data to calculate our elemental fluxes. However, the similarity
in the POC/Th ratio between the >1-54 m and >54m particles at depth during the entire experiment
and in surface waters at the beginning of SOFeX provides rather narrow constraints on the possible
23
range of POC/Th ratio for any particular time and depth, hence the magnitude of the POC flux is
tightly constrained by these data.
Filtration is an operational procedure that can introduce collection biases based upon sampling
and experimental conditions and particle type. Methods related differences have been found for
example in the concentration of POC measured using bottle and in-situ pump methods (Gardner et al.
2003; Moran et al. 1999). We contend that since we measure both POC and 234Th together on the
same filter matrix collected using a single method, the only possible methods bias would be if we had
missed a class of sinking particles with a different POC/Th ratio and these missing particles dominated
the local particle flux. Alternatively if the filtration caused the loss of DOC from cells, that loss would
have to be large and the material would have to have a dramatically different POC/Th ratio in order
significantly bias the flux estimate. It is important to note that the total range in observed POC/Th
ratios should not be used as an estimate of the accuracy of the method (Moran et al. 2003). Using this
empirical approach, it is only appropriate to apply the POC/Th ratio characteristic of the particles at the
base of any given depth integral to calculate the POC flux at that depth (Benitez-Nelson and Charette
2004). The range in observed POC/Th ratios at a given depth and time are considerably smaller than
the total range of all time series POC/Th data for all depths (Figure 8).
We calculated the flux of POC for IN and OUT stations using a fit to the profile of >54 m
POC/Th ratios assuming a linear decrease between the mixed layer and 75 m and using the average of
the 50-100 m POC/Th data below (Figure 9; Table 2). Since the IN patch POC flux is essentially the
product of Figures 7A and 9A, the flux profile for POC increases substantially both below the Fe patch
at 50 m, and in response to the higher 234Th fluxes below that more than compensate for the decrease in
POC/Th in this depth zone (Figure 7C). POC fluxes at the base of the mixed layer (50 m) of the patch
increased from 3 ± 0.9 at the start to 9.3 ± 2.3 mmol m-2 d-1 by our last observation. Flux increases
24
were largest below this layer, increasing from 1.6 ± 0.9 to 11.8 ± 3.2 mmol m-2 d-1 at 100 m from the
start to the end of the experiment (Figures 7C, Table 2). At the OUT stations, POC fluxes increased
from about 2 to 5 mmol m-2 d-1 in the mixed layer and decreased below, but did not vary much with
time (Figure 7D). The 1-3 mmol m-2 d-1 errors estimated here for the POC flux were propagated from
both the error on the 234Th flux model (which increases with depth) and variability in POC/Th data.
For the POC/Th error we use the standard deviation among >54 m POC/Th samples collected at the
depth of interest (averaging all time points since we do not see any statistically significant temporal
trend for >54 m particles); in general, this variability decreased with increasing depth (Table 2).
We do not report here the flux of PON specifically, since by using the ratio of C/N reported in
Web Appendix 2 at http://www.aslo.org/lo/toc/vol_xx/issue_x/xxxxa2.pdf, one can simply convert
from C to N units. This ratio (in moles) on the >54 m particles averages 6.0 in the upper 50 m and
6.9 between 50-100 m at the IN stations, and is similar OUT.
Biogenic silica flux
The flux of biogenic silica in the upper 125 m was estimated in a manner exactly analogous to
that described for POC. The bSiO2/Th ratio for >54 m particles in each depth horizon showed no
significant change with time, but varied by almost a factor of two between the surface and 100m (Web
Appendix 2 at http://www.aslo.org/lo/toc/vol_xx/issue_x/xxxxa2.pdf). Thus we applied a depth
varying, but temporally constant bSiO2/Th ratio to calculate bSiO2 flux from Th flux for each 25 m
depth integral from the surface to 125m (Table 2, Figure 9). The bSiO2/Th ratio for each depth integral
was calculated by using the same interpolation scheme as described for POC/Th.
The flux of bSiO2 at the base of the mixed layer (50 m) of the patch increased 3-fold during the
experiment rising from 0.2 ± 0.1 mmol Si m-2 d-1 at the start to 0.6 ± 0.3 mmol Si m-2 d-1 by our last
25
observation. As with POC flux the increase in the flux of bSiO2 was greatest below the mixed layer at
100 m where bSiO2 export increased 6.5-fold from 0.2 ± 0.1 at the start to 1.3 ± 0.6 mmol Si m-2 d-1 at
the end of the experiment (Figures 7E, Table 2). In contrast, the flux of bSiO2 did not change in the
upper 50 m at the OUT stations and increased slightly, but within measurement error, between 50 and
125 m (Figure 7F).
DISCUSSION
Comparison of export flux to SOFeX carbon budget
We can use estimates of changes in C stock as an independent check on the magnitude of the
POC export predicted here using the 234Th approach. Coale et al. (2004) looked at measurements of
TCO2, POC and gas exchange to determine a net decrease in C stocks in the South Patch mixed layer
of 7 ± 7 µmol L-1, equivalent to a flux of 12 mmol C m-2 d-1 when averaged over 50 m and 30 days.
Corrections for patch dilution were not made, which would increase these geochemical signals by a
factor of three. As seen in the large uncertainty on this C budget, it is difficult to quantify small
changes in C loss from the depletion of much larger inorganic C stocks and build up of organic C
pools. Not included in the Coale et al. (2004) C balance are changes in the DOC pool, so this change
in C stocks could be accounted for either by POC export or DOC build-up. Our average 50 m POC
flux IN and integrated over the same 30 days would average 7 mmol m-2 d-1 (Table 2). Thus both
estimates are in agreement, but the comparison is not rigorous.
Magnitude of SOFeX fluxes and comparisons to AESOPS
Despite enhanced fluxes of POC and bSiO2 in response to Fe fertilization during SOFeX, the
magnitude of these fluxes was not particularly large relative to the natural cycle of C uptake and export
26
at this site. For example, at this same region and season in 1998 (US Joint Global Ocean Flux Study,
Antarctic Ecosystem Southern Ocean Process Study- AESOPS), we measured significantly lower 234Th
activities and higher associated particulate 234Th, POC and bSiO2 fluxes, even when compared to the
fertilized patch after 28 days (Table 3). The AESOPS POC fluxes were 5-10 times greater than the
SOFeX OUT stations, and the final SOFeX IN POC flux, was only half that observed in 1998. Also,
there were 5 times greater bSiO2 fluxes in February 1998 (bSiO2:C export ratio = 0.33) than at the end
of SOFeX IN (Table 3).
It is likely that the lower export fluxes and lower bSiO2:C export ratios during SOFeX relative
to AESOPS reflect a lower abundance of diatoms both in the ambient waters and in the plankton
assemblage arising in response to Fe. This is supported by the significantly lower bSiO2
concentrations measured during SOFeX compared to February 1998 (Table 3). There were also
differences in the physical setting as well. The AESOPS stations had similar salinities and a somewhat
shallower mixed layer in January, but AESOPS temperatures were about 1°C warmer within the mixed
layer and the mid January biomass and primary production levels were much higher than equivalent
SOFeX OUT stations. Thus, there were large differences in the biogeochemical conditions at the start
of SOFeX that may explain the low export fluxes (colder waters, lower diatom abundances). This
comparison also points to the difficulty in characterizing Southern Ocean biogeochemistry, when even
within a relatively well studied region- in this case, south of ACC front within the seasonal ice zone
and in permanently high dissolved Si and N waters- there can be considerable interannual variability in
production and export fluxes.
With respect to the SOFeX fluxes, it is important to note that the effects of iron addition very
likely continued beyond the occupation of our last station. On day 28 surface waters were still Fereplete relative to phytoplankton growth requirements, such that biomass, photosynthetic efficiency,
27
and other indicators suggest that the bloom had not ended (Coale et al. 2004). During AESOPS, a lag
time of two weeks to >1 month was observed between the onset and termination of the bloom (Abbott
et al. 2000; Buesseler et al. 2003). Here Fe was added four times through day 12 in a deliberate
attempt to keep the bloom from becoming Fe limited. Thus, our observation period was not likely long
enough to see the onset of Fe limitation or other factors that might cause the termination of the SOFeX
bloom.
Rates of C and Si export vs. production
Prior studies of Southern Ocean biogeochemistry suggest particularly high rates of export on
sinking particles relative to primary production or bSiO2 uptake south of the Antarctic Polar Front
(Buesseler et al., 2001; 2003). During SOFeX, the efficiency of POC and bSiO2 export, defined as the
ratio of export to production, was calculated for weeks 1, 2 and 3 using all available data integrated
over the upper 50m. While primary production, Si uptake and POC export all increased during SOFeX
in response to Fe, there was a significant difference between the export efficiency of organic carbon as
compared to bSiO2. Both tended to be higher at OUT stations compared to IN stations, but the
differences were generally insignificant for bSiO2. For the C export efficiency, the lower fraction of C
lost on sinking particles IN may be tied to Fe replete conditions favoring more efficient retention of
POC in response to the Fe induced diatom bloom (see below).
Relative to each other, the efficiency of POC export was higher than that for biogenic silica
both IN and OUT (above 1:1 line in Figure 10) with the ratio of POC: bSiO2 export efficiency varying
between 1.3 and 2.4. If the source of exported POC was entirely diatoms, then the greater efficiency of
POC export compared to that of bSiO2 would imply that bSiO2 was more effectively recycled in the
upper 50 m than was organic matter. This is unlikely as the turnover of organic matter is typically
28
more rapid than that of bSiO2 in surface waters. A more plausible explanation is that non siliceous
plankton were preferentially being exported over diatoms both in and out of the patch. This was
unexpected as diatoms dominated the increase in phytoplankton biomass within the patch (Coale et al.,
2004). Export of this bloom would yield the same export efficiency for carbon and silica in the patch.
The fact that this did not occur argues that the Fe-induced diatom bloom was not a dominant
component of carbon export during the time frame of our experiment. Efficient retention of the diatom
bloom in the surface waters is consistent with improved buoyancy control that would arise after release
from Fe stress. Improved buoyancy control is observed under nutrient-replete conditions for many
phytoplankton (Bienfang et al. 1982; Smayda 1970), but in this case the diatoms were the major
benefactors of the release from Fe stress. Whatever the mechanism retaining the bloom in surface
waters, there was a large potential for carbon export by diatoms that was unrealized within the
timeframe of our experiment.
Comparison to other Southern Ocean Fe studies
There have been only three other Southern Ocean iron fertilization experiments to date that we
can compare to the SOFeX South Patch: SOIREE (Southern Ocean Iron Enrichment Experiment),
EisenEx-1 (Eisen (=Iron) Experiment), and the North Patch of SOFeX. All three produced notable
increases in biomass and associated decreases in dissolved inorganic carbon and associated nutrients
(Boyd et al. 1999; Coale et al. 2004; Smetacek 2001; Bishop et al. 2004). SOIREE was initiated in
February 1999 at 60oS, 140oE (Boyd et al. 1999), and during a 13 day observation period, the
234
Th:238U ratio actually increased similarly at both IN and OUT stations. The slope of the increase was
explained by the natural ingrowth of 234Th from 238U during conditions of near zero net particle flux
(Charette and Buesseler 2000). The 234Th results agreed with the lack of clear IN vs. OUT differences
29
in shallow sediment trap flux (Nodder et al. 2001) and the satellite observation of the persistence of the
SOIREE patch 55 days after the beginning of the experiment (Abraham et al. 2000). Also, a
comprehensive carbon balance for SOIREE did not reveal a significant carbon deficit that would
indicate POC export (Bakker et al. 2001).
The pre-Fe release condition during EisenEx differed from SOIREE and SOFeX in that at the
start of observations (November 2000; 48oS, 21oE), there were no measurable 234Th disequilibria
(234Th = 238U; M. Rutgers van der Loeff, personal communication). These data indicate the absence of
significant particle export prior to arrival, and therefore EisenEx began prior to the start of the natural
flux cycle. Indeed, as the 22 day experiment progressed, 234Th activities decreased indicating particle
export, but there was no difference between IN and OUT stations (U. Riebesell and M. M. Rutgers van
der Loeff, personal communication). Thus, based on the 234Th data alone, it does not appear as if there
was Fe-enhanced particle export relative to the natural phytoplankton dynamics.
Somewhat at odds with the EisenEx 234Th data were attempts to balance carbon stocks. Based
upon the difference in dissolved inorganic C and POC stocks alone, Riebessel et al. (2003) conclude
that export IN was greater than OUT by 1000 mmol C m-2 or 48 mmol C m-2 d-1 for the upper 80 m. As
with the SOFeX it is very clear that biomass responded to Fe, but an exact accounting for export by
looking at the difference in C stocks is difficult to constrain due in part to the lack of DOC data.
The SOFeX North Patch (January-February, 2002; 56oS 172oW) attained roughly twice the
primary production rates, but lower Chl a levels than in the South patch after a similar series of Fe
additions (Coale et al. 2004). Ship based observations were very limited to early and late conditions
for the North Patch, but based upon data from an autonomous profiling float that measured particle
abundances continuously, Bishop et al. (2004) conclude that there was a significant POC flux increase
during a period 25-45 days after the first Fe addition. Using a POC flux calibration from a similar float
30
in the South Patch, they estimate a POC flux of 2.4 to 23 mmol C m-2 d-1 at 100 m when averaged over
the entire 50 day period. They favor the upper range of these POC flux estimates and assuming this
flux calibration is correct, the POC fluxes would be significantly higher than the 7 mmol C m-2 d-1 we
observed on average during the first 28 days of the South Patch.
Coale et al. (2004) estimated C balance for the North Patch using methods similar to those
employed for the South Patch equivalent to 8 mmol C m-2 d-1 (both after 28-30 days). However, due to
limited sampling in the North Patch and a lack of data for some of the major C reservoirs, in particular
dissolved organic carbon, the errors on these C balances are large. Bishop et al. (2004) speculate that
the export response in the North Patch was triggered by a subduction of POC due to a change in
stratification that isolated the iron-stimulated biomass in a layer with reduced light and lower
turbulence, however the lack of concomitant biogeochemical data makes it impossible to rule out other
causes. A similar subduction event was not observed in the South Patch.
We observed changes in the distribution of the particle reactive radionuclide 234Th in response
to Fe addition in HNLC waters of the Southern Ocean. The Lagrangian sampling of SF6 tagged waters
allowed for an ideal application of this tracer, as we could follow the evolving 234Th activity
distribution and quantify horizontal and vertical exchange processes to resolve small changes in 234Th
fluxes. The main feature is a decrease in 234Th activity within the mixed layer attributable to enhanced
particle export in response to iron.
We also saw a decrease in excess 234Th below the Fe fertilized patch, which indicated
decreased particle remineralization beneath the patch. This implies that the particles leaving the upper
ocean are more robust (less labile or sinking more rapidly) and/or the biological and physical processes
responsible for particle degradation and break-up below the mixed layer have become less efficient.
31
This subsurface response is not currently considered in our understanding of the possible
biogeochemical impacts of Fe in the ocean.
When converted to POC and bSiO2 export rates, these fluxes were not large relative to natural
blooms, and were in fact smaller than those observed under natural conditions at this site during a
different field season. The lack of a large SOFeX export response is consistent with models suggesting
that at the 10-100 km patch scales of SOFeX, natural dilution effects would reduce cell numbers and
thus limit aggregation and hence the export response (Boyd et al. 2002). We can only speculate as to
the ultimate fate of C in the SOFeX South patch, as no further ship observations or satellite images
were obtained after the first month.
Ultimately, it is the sequestration of C and associated elements to the deep ocean that has the
potential to influence climate as put forward by the iron hypothesis. Unfortunately, C flux to the deep
ocean is difficult to assess. Ideally, one would quantify not just what leaves the Fe fertilized patch, but
subsurface effects, such as seen here in the remineralization patterns of sinking particles and controls
thereof. Mass balance approaches to quantify surface ocean C losses are inherently difficult, due to the
dynamic nature of patch dilution and efforts needed to quantify in 4-D all inorganic and organic C
forms and exchanges, especially since the export fluxes are much smaller than standing stocks. Part of
this difficulty is overcome using 234Th as an in-situ tracer, as transport terms are small relative to the
radioactive production and decay rates which are quantified by the 234Th:238U disequilibrium and
dominate the Th export flux calculation.
Many questions remain concerning the longer-term fate of iron induced blooms, including what
factors trigger bloom termination, and what fraction of the C taken up during photosynthesis would
reach deeper waters that are not regularly ventilated. To better examine these questions we must
design experiments that stimulate substantial biomass increase, and also terminate Fe addition early
32
enough in the experiment so that we can observe the return of Fe limitation. Measurement of the
export of multiple elements provides a means to assess the export contribution of different
phytoplankton groups. The comparison of flux to rates of production yields insights into relative
recycling rates of organic and inorganic constituents. Also, dilution effects need to be considered not
just in mass balancing of major elements, but in the dynamics of aggregation. Larger patch scales may
enhance aggregation and provide conditions more like the large-scale changes attributed to shifts in
dust-born Fe input associated with glacial and interglacial cycles.
Thus far, there have been very few data collected on POC export response to Fe. This study is
one of the most complete, and suggests relatively modest export response and low efficiency of C
export relative to Fe added (Buesseler and Boyd 2003). The assumption that Fe addition would lead
directly to large export has proven difficult to demonstrate. Studies to date even outside of the
Southern Ocean are not convincing in demonstrating a large export response. There are limitations
with different methods and difficulties in constraining C stocks and fluxes over perhaps too short an
observation period. However, what has been found in the Southern Ocean supports a response that
includes efficient recycling of Fe and observations of ecosystems that maintain high photosynthetic
efficiency and chlorophyll biomass many weeks after Fe addition. This was demonstrated most
dramatically by the satellite observations in SOIREE of the surface Chl a feature almost 2 months after
Fe addition (Abraham et al. 2000). Therefore the question remains open as to how the response to Fe
as seen in enhanced phytoplankton growth in HNLC regions, can be extrapolated to export, and how
this might change under different conditions. There is certainly no direct support for geoengineering
proposals to fertilize the ocean which claim to reduce buildup of greenhouse CO2 via C sequestration
in the deep ocean. However, since past changes in atmospheric CO2 and climate have been tightly
33
correlated to dust and hence Fe sources to the ocean (Martin 1990), the issue of the ocean C flux
response to iron and the fate of this sinking C remains an important question to resolve.
34
REFERENCES
Abbott, M. R., J. G. Richman, R. M. Letelier, and J. S. Bartlett. 2000. The spring bloom in the
Antarctic Polar Frontal Zone as observed from a mesoscale array of bio-optical sensors. DeepSea Research II 47(15-16): 3285-3314.
Abraham, E. R., C. S. Law, P. W. Boyd, S. J. Lavender, M. T. Maldonado, and A. R. Bowie. 2000.
Importance of stirring in the development of an iron-fertilized phytoplankton bloom. Nature
407: 727-730.
Bakker, D. C. E., A. J. Watson, and C. S. Law. 2001. Southern Ocean iron enrichment promotes
inorganic carbon drawdown. Deep Sea Res. II 48: 2483-2508.
Barber, R. T. and others 1996. Primary productivity and its regulation in the equatorial Pacific during
and following the 1991-1992 El Nino. Deep-Sea Research Part II-Topical Studies in
Oceanography 43: 933-969.
Benitez-Nelson, C., K. O. Buesseler, and G. Crossin. 2000. Upper ocean carbon export, horizontal
transport, and vertical eddy diffusivity in the southwestern Gulf of Maine. Continental Shelf
Research 20: 707-736.
Benitez-Nelson, C. and others 2001. Testing a new small-volume technique for determining thorium234 in seawater. Journal of Radioanalytical and Nuclear Chemistry 248(3): 795-799.
Benitez-Nelson, C. R., and M. A. Charette. 2004. Uncertainty versus variability in upper ocean carbon
flux estimates. Limnol. Oceanogr. 49(4): 1218-1220.
Bidigare, R. R. and others 1999. Iron-stimulated changes in 13C fractionation and export by equatorial
Pacific phytoplankton: Toward a paleogrowth rate proxy. Paleoceanography 14(5): 589-595.
Bishop, J. K. B., T. J. Wood, R. E. Davis, and J. T. Sherman. 2004. Robotic Observations of Enhanced
Carbon Biomass and Export at 55S During SOFeX. Science 304: 417-420.
35
Bowie, A. R. and others 2001. The fate of added iron during a mesoscale fertilisation experiment in the
Southern Ocean. Deep-Sea Research II 48 (11-12): 2703-2743.
Boyd, P. W. and others 1999. A mesoscale phytoplankton bloom in the polar Southern Ocean
stimulated by iron fertilization. Nature 407: 695-702.
Boyd, P. W., and E. R. Abraham. 2001. Iron-mediated changes in phytoplankton photosynthetic
competence during SOIREE. Deep-Sea Research II 48(11-12): 2529-2550.
Boyd, P. W., G. A. Jackson, and A. M. Waite. 2002. Are mesoscale perturbation experiments in polar
waters prone to physical artefacts? Evidence from algal aggregation modelling studies.
Geophysical Research Letters 29: 10.1029/2001GL014210.
Brzezinski, M. A., D. M. Nelson, V. M. Franck, and D. E. Sigmon. 2001. Silicon dynamics within an
intense open-ocean diatom bloom in the Pacific sector of the Southern Ocean. Deep-Sea
Research II 48(19-20): 3997-4018.
Buesseler, K. O., M. P. Bacon, J. K. Cochran, and H. D. Livingston. 1992. Carbon and nitrogen export
during the JGOFS North Atlantic Bloom Experiment estimated from 234Th: 238U disequilibria.
Deep-Sea Research I 39: 1115-1137.
Buesseler, K. O., J. A. Andrews, M. C. Hartman, R. Belastock, and F. Chai. 1995. Regional estimates
of the export flux of particulate organic carbon derived from thorium-234 during the JGOFS
EQPAC program. Deep-Sea Research II 42: 777-804.
Buesseler, K. O. 1998. The decoupling of production and particulate export in the surface ocean.
Global Biogechemical Cycles 12: 297-310.
Buesseler, K. O. and others 1998. Upper ocean export of particulate organic carbon in the Arabian Sea
derived from thorium-234. Deep-Sea Research II 45(10-11): 2461-2487.
36
Buesseler, K. O. and others 2001a. An intercomparison of small- and large-volume techniques for
thorium-234 in seawater. Marine Chemistry 74(1): 15-28.
Buesseler, K. O. and others. 2001b. Upper ocean export of particulate organic carbon and biogenic
silica in the Southern Ocean along 170°W. Deep-Sea Research II 48: 4275-4297.
Buesseler, K. O., and P. W. Boyd. 2003. Will ocean fertilization work? Science 300: 67-68.
Buesseler, K. O., R. T. Barber, M.-L. Dickson, M. R. Hiscock, J. K. Moore, and R. Sambrotto. 2003.
The effect of marginal ice-edge dynamics on production and export in the Southern Ocean
along 170ºW. Deep-Sea Research II 50(3-4): 579-603.
Buesseler, K. O., J. E. Andrews, S. M. Pike, and M. A. Charette. 2004. The effects of iron fertilization
on carbon sequestration in the Southern Ocean. Science 304: 414-417.
Bienfang, P. K., P. J. Harrison and L. M. Quarmby. 1982. Sinking rate response to depletion of nitrate,
phosphate and silicate in four marine diatoms. Mar. Biol. 67: 295-302.
Charette, M. A., and K. O. Buesseler. 2000. Does iron fertilization lead to rapid carbon export in the
Southern Ocean? Geochemistry, Geophysics, Geosystems: Paper number 2000GC000069.
Charette, M. A., S. B. Moran, S. M. Pike, and J. N. Smith. 2001. Investigating the carbon cycle in the
Gulf of Maine using the natural tracer thorium-234. Journal of Geophysical Research-Oceans
106: 11553-11579.
Chen, J. H., R. L. Edwards, and G. J. Wasserburg. 1986. 238U, 234U and 232Th in seawater. Earth and
Planetary Science Letters 80: 241-251.
Coale, K. H. and others 1996. A massive phytoplankton bloom induced by an ecosystem-scale iron
fertilization experiment in the equatorial Pacific Ocean. Nature 383: 495-501.
37
Coale, K. H., K. S. Johnson, S. E. Fitzwater, S. P. G. Blain, T. P. Stanton, and T. L. Coley. 1998. Iron
Ex-I, an in situ iron-enrichment experiment: Experimental design, implementation and results.
Deep Sea Res. II 45: 919-946.
Coale, K. H. and others 2004. Southern Ocean Iron Enrichment Experiments (SOFeX) Expand Extent
of Iron Limitation and Carbon Flux in Southern Ocean. Science 304: 408-413.
Cochran, J. K., K. A. Roberts, C. Barnes, and D. Achman. 1997. Radionuclides as indicators of particle
and carbon dynamics on the East Greenland Shelf, p. 129-136. In J. C. G. P. Germain, P.
Guegueniat and H. Metivier [ed.], RADOC 96-97 "Radionuclides in the Oceans".
Radioprotection- colloques, 32 (C2).
Dunne, J. P and J. W. Murray. 1999. Sensitivity of 234Th export to physical processes in the central
equatorial Pacific. Deep Sea Res. I 46(5): 831-854.
Gardner, W. D., M. J. Richardson, C. A. Carlson, D. A. Hansell, and A. V. Mishonov. 2003.
Determining true particulate organic carbon: bottles, pumps and methodologies. Deep Sea Res.
II 50: 655-692.
Goldson, L. E. 2004. Vertical mixing across the seasonal pycnocline in the Southern Ocean: Studies
using SF6 tracer. PhD Thesis. University of East Anglia, UK.
Gustafsson, Ö., K. O. Buesseler, W. R. Geyer, S. B. Moran, and P. M. Gschwend. 1998. On the
relative significance of horizontal and vertical transport of chemicals in the coastal ocean:
Application of a two-dimensional Th-234 cycling model. Continental Shelf Research 18: 805829.
Holm-Hansen, O., C. J. Lorenzen, R. W. Holmes, and J. D. H. Strickland. 1965. Fluorometric
determination of chlorophyll. Journal du Conseil 30: 3-15.
38
Hunter, C. N., R. M. Gordon, S. E. Fitzwater, and K. H. Coale. 1996. A rosette system for the
collection of trace metal clean seawater. Limnology and Oceanography 41: 1367-1372.
Huntsman, S. A., and R. T. Barber. 1977. Primary production off northwest Africa: the relationship to
wind and nutrient conditions. Deep Sea Research 24: 25-33.
Lorenzen, C. J. 1966. A method for the continuous measurement of in vivo chlorophyll concentration.
Deep Sea Research 13: 223-227.
Martin, J. H. 1990. Glacial-Interglacial CO2 Change: the Iron Hypothesis. Paleoceanography 5: 1-13.
Moran, S. B., M. A. Charette, S. M. Pike, and C. A. Wicklund. 1999. Differences in seawater
particulate organic carbon concentration in samples collected using small-volume and largevolume methods: the importance of DOC adsorption to the filter blank. Marine Chemistry 67:
33-42.
Moran, S. B. and others 2003. Does 234Th/238U disequilibrium provide an accurate record of the export
flux of particulate organic carbon from the upper ocean? Limnology and Oceanography 48(3):
1018-1029.
Morrison, J. M. and others 2001. Seasonal evolution of hydrographic properties in the Antarctic
Circumpolar Current at 170ºW during 1997-1998. Deep-Sea Research II 48(19-20): 39433972.
Nodder, S. D. and others 2001. Particle transformations and export flux during an in situ ironstimulated algal bloom in the Southern Ocean. Geophysical Research Letters 28(12): 24092412.
Pike, S. M., and S. B. Moran. 1997. Use of Poretics 0.7 µm pore size glass fiber filters for
determination of particulate organic carbon and nitrogen in seawater and freshwater. Mar.
Chem. 57: 355-360.
39
Pike, S., K. O. Buesseler, J. A. Andrews, and N. Savoye. 2004. Quantification of 234Th recovery in
small volume sea water samples by inductively coupled plasma mass spectrometry. Journal of
Radioanalytical and Nuclear Chemistry, in press.
Santschi, P. H. and others 2003. Control of acid polysaccharide production and 234Th and POC export
fluxes by marine organisms. Geophysical Research Letters 30: 1044,
doi:1010.1029/2002GL016046.
Smayda, T. J. 1970. The suspension and sinking of phytoplankton in the sea. Oceanogr. Mar. Biol. An.
Rev. 8: 353-414.
Smetacek, V. S. 2001. U.S. JGOFS Newsletter 11: 11.
Tsunogai, S., K. Taguchi, and K. Harada. 1986. Seasonal variation in the difference between observed
and calculated particulate fluxes of Th-234 in Funka Bay, Japan. J. Oceanogr. Soc. Japan 42:
91-98.
Venrick, E. L., and T. L. Hayward. 1984. Determining chlorophyll on the 1984 CalCOFI surveys.
California Cooperative Oceanic Fisheries Investigations Reports 25: 74-79.
Watson, A. J., P. J. Liss, and R. A. Duce. 1991. Design of a small-scale in situ iron fertilization
experiment. Limnol. Oceanogr. 36: 1960-1965.
Wei, C.-L., and J. W. Murray. 1992. Temporal variations of 234Th activity in the water column of
Dabob Bay: Particle scavenging. Limnology and Oceanography 37: 296-314.
40
Web Appendix 1: South patch total 234Th data.
Depth Total 234Th
Cast ID
238
U
Temp
(m) (dpm L-1) ±error (dpm L-1) (°C)
Salinity
5.0
14.0
66.34°S 171.96°W 22.0
33.0
30Jan02
0129 h GMT
42.0
Fe-Add = 5.75d
54.0
84.0
126.0
2.10
2.03
1.98
1.97
2.23
2.29
2.56
2.48
±0.05
±0.05
±0.05
±0.05
±0.05
±0.05
±0.05
±0.05
2.38
2.38
2.38
2.38
2.38
2.40
2.41
2.41
-0.499
-0.518
-0.506
-0.802
-1.680
-1.706
-1.438
0.760
33.721
33.806
33.757
33.867
34.162
34.207
34.301
34.573
2.3
16.9
66.31°S 171.39°W 30.0
43.6
31Jan02
0307 h GMT
53.4
Fe-Add = 6.82d
69.0
108.0
162.7
1.85
1.94
1.81
1.94
2.04
2.51
2.72
2.55
±0.05
±0.05
±0.05
±0.05
±0.05
±0.05
±0.05
±0.05
2.37
2.38
2.38
2.38
2.37
2.38
2.40
2.44
-0.672
-0.670
-0.681
-1.371
-1.719
-1.763
-0.825
1.199
33.815
33.794
33.778
34.167
34.207
34.260
34.367
34.653
4.4
13.2
21.6
32.6
42.4
53.3
83.1
2.12
2.03
2.04
1.91
1.93
2.17
2.63
±0.04
±0.04
±0.04
±0.04
±0.04
±0.05
±0.05
2.38
2.38
2.38
2.38
2.38
2.40
2.41
-0.740
-0.764
-0.768
-0.773
-1.010
-1.708
-1.640
33.759
33.757
33.744
33.746
33.908
34.152
34.275
6.0
11.3
17.9
26.0
34.3
44.7
71.1
1.91
1.88
1.90
1.92
1.97
2.02
2.49
±0.04
±0.03
±0.04
±0.04
±0.04
±0.04
±0.05
2.38
2.39
2.38
2.38
2.38
2.38
2.41
-0.699
-0.691
-0.705
-0.683
-0.678
-0.794
-1.694
33.917
33.886
33.964
33.850
33.831
33.892
34.329
5.0
12.2
66.31°S 171.39°W 21.7
34.3
4Feb02
2225 h GMT
43.7
Fe-Add = 12.62d
55.2
84.8
129.6
1.85
1.79
1.83
1.71
1.78
2.08
2.59
2.53
±0.04
±0.04
±0.04
±0.04
±0.04
±0.04
±0.04
±0.04
2.37
2.38
2.38
2.38
2.37
2.38
2.40
2.44
-0.495
-0.543
-0.557
-0.583
-0.655
-1.575
-1.034
1.191
33.353
33.578
33.230
33.797
33.802
34.153
34.310
34.593
Mel TM 08
IN
Mel TM 11
OUT
Mel TM 17
OUT
66.59°S 172.20°W
1Feb02
2051 h GMT
Fe-Add = 8.56d
Mel TM 28
IN
66.26°S 171.96°W
4Feb02
0137 h GMT
Fe-Add = 10.75d
Mel TM 33
OUT
Cast ID
Mel TM 37
IN
66.17°S 172.21°W
8Feb02
2012 h GMT
Fe-Add = 15.53d
Mel TM 43
IN
66.10°S 172.07°W
10Feb02
2044 h GMT
Fe-Add = 17.55d
Mel TM 46
OUT
66.03°S 170.99°W
12Feb02
2100 h GMT
Fe-Add = 19.56d
Mel TM 52
IN
65.95°S 172.19°W
14Feb02
2224 h GMT
Fe-Add = 21.62d
PS BC 01
IN
65.90°S 171.97°W
15Feb02
0010 h GMT
Fe-Add = 21.69d
Depth
Total
234
Th
5.1
11.1
17.0
27.0
32.5
42.9
67.2
100.9
1.98
2.02
1.99
1.96
1.82
1.98
2.36
2.44
±0.04
±0.04
±0.04
±0.04
±0.04
±0.04
±0.04
±0.04
2.39
2.38
2.39
2.39
2.38
2.39
2.41
2.42
-0.646
-0.643
-0.665
-0.645
-0.648
-0.764
-1.309
-0.487
33.881
33.866
33.929
33.854
33.840
34.006
34.265
34.375
3.6
8.1
13.7
19.6
26.0
32.7
76.8
1.91
1.91
1.83
1.76
1.87
1.81
2.64
±0.05
±0.04
±0.04
±0.04
±0.04
±0.05
±0.05
2.35
2.33
2.33
2.38
2.37
2.39
2.41
-0.348
-0.441
-0.484
-0.520
-0.560
-0.641
-1.285
33.134
33.432
33.656
33.757
33.813
33.938
34.286
3.7
12.9
21.6
32.5
40.6
53.9
83.0
123.7
1.99
1.88
1.99
1.84
1.97
2.12
2.61
2.47
±0.04
±0.04
±0.05
±0.04
±0.04
±0.05
±0.05
±0.05
2.35
2.35
2.35
2.36
2.35
2.37
2.38
2.41
-0.382
-0.392
-0.394
-0.451
-0.459
-0.938
-1.571
0.636
33.870
33.869
33.869
33.871
33.870
33.995
34.272
34.590
2.8
8.3
12.7
19.6
24.6
31.0
50.9
75.8
1.96
1.99
1.99
1.81
1.99
1.82
1.98
2.27
±0.04
±0.04
±0.04
±0.04
±0.04
±0.03
±0.03
±0.04
2.36
2.36
2.37
2.36
2.36
2.36
2.36
2.38
-0.370
-0.370
-0.372
-0.372
-0.426
-0.592
-0.900
-1.014
33.917
33.917
33.917
33.917
33.918
33.921
34.009
34.330
33.9
50.9
68.0
1.76
1.83
1.95
±0.03
±0.04
±0.03
2.39
2.39
2.39
238
U Temp
(dpm L1
(m) (dpm L-1)±error
)
(°C) Salinity
nd 33.911
nd 33.905
nd 33.952
41
Web Appendix 1: South patch total 234Th data, continued.
Depth Total 234Th
Cast ID
PS BC 02
IN
65.90°S 171.97°W
15Feb02
0235 h GMT
Fe-Add = 21.80d
PS CTD 02
IN
65.96°S 172.07°W
16Feb02
2235 h GMT
Fe-Add = 23.63d
PS BC 03
IN
65.89°S 172.13°W
17Feb02
0345 h GMT
Fe-Add = 23.84d
PS BC 04
OUT
66.17°S 172.61°W
17Feb02
2330 h GMT
Fe-Add = 24.67d
238
U
-1
Temp
Depth
-1
(m) (dpm L ) ±error (dpm L ) (°C) Salinity
28.8
1.83
±0.05
2.39
nd 33.893
48.1
1.97
±0.04
2.39
nd 33.905
67.3
2.17
±0.04
2.40
nd 34.091
86.5
2.34
±0.04
2.41
nd 34.239
105.7
2.34
±0.04
2.42
nd 34.367
125.0
2.61
±0.04
2.43
nd 34.555
Cast ID
PS BC 05
OUT
10.6
20.3
50.5
70.0
79.6
100.5
149.2
1.74
1.68
1.98
2.23
2.36
2.36
2.46
±0.03
±0.03
±0.04
±0.05
±0.05
±0.04
±0.04
2.39
2.39
2.40
2.41
2.42
2.43
2.44
-0.382
-0.380
-0.853
-1.257
-0.916
0.307
1.327
33.918
33.934
34.010
34.217
34.336
34.533
34.694
PS CTD 04
IN
19.0
39.0
59.0
79.0
99.0
119.0
1.82
1.80
1.74
2.22
2.54
2.51
±0.03
±0.05
±0.03
±0.04
±0.04
±0.04
2.39
2.39
2.39
2.41
2.43
2.44
-0.379
-0.427
-0.898
-1.349
-0.382
0.769
33.942
33.980
34.019
34.247
34.441
34.604
32.4
52.4
72.4
92.4
112.4
132.4
2.11
1.92
2.19
2.58
2.39
2.49
±0.04
±0.03
±0.04
±0.04
±0.06
±0.04
2.38
2.38
2.40
2.42
2.42
2.44
-0.356
-0.497
-1.064
-1.244
-0.167
0.797
33.766
33.837
33.967
34.276
34.439
34.588
65.57°S 172.93°W
18Feb02
0452 h GMT
Fe-Add = 24.89d
65.91°S 172.19°W
19Feb02
0615 h GMT
Fe-Add = 25.95d
PS CTD 05
OUT
65.91°S 170.79°W
19Feb02
2205 h GMT
Fe-Add = 26.61d
PS CTD 06
IN
65.85°S 171.95°W
20Feb02
1753 h GMT
Fe-Add = 27.43d
Total
234
Th
238
U
-1
Temp
-1
(m) (dpm L ) ±error (dpm L ) (°C) Salinity
6.6 1.94
±0.03
2.38
-0.402 33.735
26.6 1.82
±0.03
2.38
-0.418 33.732
46.6 1.91
±0.03
2.38
-0.446 33.743
66.6 2.26
±0.04
2.40
-1.268 34.071
86.6 2.41
±0.03
2.42
-0.234 34.433
106.6 2.56
±0.04
2.43
0.549 34.454
20.7
49.9
70.3
125.3
1.72
2.12
2.69
±0.04
±0.06
±0.04
±0.04
2.39
2.39
2.40
2.44
-0.276
-0.384
-1.176
0.749
33.904
33.913
34.096
34.602
19.6
50.0
69.9
100.0
125.2
1.92
1.93
2.39
2.61
2.34
±0.04
±0.04
±0.05
±0.05
±0.06
2.39
2.39
2.41
2.43
2.44
-0.231
-0.444
-1.270
-0.429
0.893
33.877
33.921
34.180
34.445
34.614
9.9
35.1
49.7
70.0
80.2
100.6
124.7
151.2
1.84
1.90
1.84
1.90
2.33
2.42
2.35
2.52
±0.04
±0.04
±0.04
±0.04
±0.05
±0.05
±0.05
±0.05
2.39
2.39
2.39
2.40
2.41
2.43
2.44
2.44
-0.275
-0.268
-0.324
-1.020
-1.374
-0.101
1.017
1.324
33.898
33.882
33.910
34.056
34.253
34.477
34.647
34.703
Mel TM # = RV Melville trace metal cast #
PS CTD # = RIB Polar Star CTD cast #
PC BC # = RIB Polar Star Bottle cast #
43
Table 1. Time series total 234Th
Slope
Depth
range
Th/t
% error
r2
0-25 m
0-50 m
0-75 m
0-100 m
0-125 m
-0.010
-0.011
-0.015
-0.014
-0.012
35
25
15
16
9
0.54
0.69
0.86
0.85
0.98
44
Web Appendix 2. South patch particle summary
234
Lat.dec
(S)
Long.dec
(W)
Date
(2002)
234
Th
Th
POC
POC
C/N
POC/Th
POC/Th
bSiO2
bSi/Th
Time Depth >54 μm
1-54 μm
>54 μm
1-54 μm molar >54 μm
1-54 μm >54 μm >54 μm
(GMT) (m) (dpm L-1) ±error (dpm L-1) ±error (μmol L-1) ±error (μmol L-1) >54 μm (μmol dpm-1) ±error (μmol dpm-1) (μmol L-1) (μmol dpm-1)
0-50m – IN
66.34
171.96
66.25
171.96
66.20
172.09
66.10
172.07
66.00
172.08
65.95
172.19
66.91
171.97
65.90
172.22
65.90
172.15
65.90
172.15
65.85
171.95
29Jan
04Feb
09Feb
11Feb
14Feb
15Feb
15Feb
17Feb
19Feb
19Feb
20Feb
2330
0245
0030
0020
2122
0030
0439
0315
0750
1054
2252
20
18
19
20
13
18
48
17
16
42
10
0.086
0.044
0.068
0.015
0.035
0.057
0.052
0.058
0.094
0.140
0.207
±0.001
±0.001
±0.001
±0.001
±0.001
±0.001
±0.000
±0.000
±0.002
±0.002
±0.001
0.352
0.318
0.340
0.275
0.292
0.169
0.211
0.173
0.218
0.112
0.261
±0.004
±0.004
±0.033
±0.005
±0.003
±0.002
±0.002
±0.004
±0.004
±0.002
±0.003
0.903
0.403
0.672
0.220
0.297
0.649
0.364
0.534
0.733
1.200
2.534
±0.101
±0.006
±0.027
±0.017
±0.015
±0.136
±0.032
±0.189
±0.082
±0.125
±0.134
2.9
2.9
3.4
4.0
3.9
5.1
2.7
3.4
6.8
3.2
4.7
8.4
6.9
6.8
5.6
6.5
7.0
3.3
3.8
5.9
5.9
5.6
10.5
9.2
9.9
14.3
8.4
11.5
7.1
9.1
7.8
8.6
12.2
±1.2
±0.2
±0.4
±1.3
±0.5
±2.4
±0.6
±3.2
±0.9
±0.9
±0.7
8.3
9.2
10.1
14.6
13.5
30.4
12.9
19.9
31.2
28.2
18.1
0.0751
0.0284
0.0344
0.0284
0.0218
0.0647
0.0550
0.0300
0.0552
0.87
0.65
0.51
1.84
0.62
1.14
1.06
0.51
0.59
0.1986
0.96
50-100m – IN
66.34
171.96
66.25
171.96
66.20
172.09
66.10
172.07
66.00
172.08
65.95
172.19
65.90
172.22
65.90
172.15
65.83
171.96
29Jan
04Feb
09Feb
11Feb
14Feb
15Feb
17Feb
19Feb
20Feb
2330
0245
0030
0020
2122
0030
0315
0750
1942
55
66
67
78
68
72
72
71
72
0.349
0.007
0.005
0.005
0.009
0.008
0.000
0.036
0.074
±0.002
±0.000
±0.001
±0.001
±0.000
±0.000
±0.000
±0.001
±0.001
0.519
0.214
0.212
0.219
0.487
0.137
0.032
0.160
0.286
±0.005
±0.003
±0.004
±0.004
±0.003
±0.013
±0.001
±0.001
±0.002
1.566
0.038
0.019a
0.023a
0.040
0.029
bd
0.108
0.404
±0.201
±0.010
±0.007
±0.006
±0.006
±0.005
2.4
0.8
0.7
0.8
1.6
0.7
4.5
5.7
3.8
5.0
4.5
3.7
±0.6
±1.5
±1.4
±1.4
±0.7
±0.7
4.6
3.7
3.3
3.7
3.3
5.0
±0.005
±0.084
1.6
1.8
7.0
bd
bd
bd
5.9
7.4
bd
7.6
6.5
3.0
5.4
±0.1
±1.1
9.7
6.3
0.1497
0.0013
0.0026
0.0027
0.0039
0.0049
0.0004
0.0139
0.0504
0.43
0.19
0.50
0.59
0.44
0.61
0.90
0.38
0.68
>100m – IN
66.91
171.97
65.90
172.15
65.85
171.95
15Feb
19Feb
20Feb
0439
1054
2252
123
117
115
0.007 ±0.000 0.173 ±0.002 0.027 ±0.005
0.005 ±0.000 0.045 ±0.001 0.011 ±0.002
0.007 ±0.000 0.144 ±0.001 0.023 ±0.004
0.4
0.2
0.5
5.9
bd
bd
4.2
2.1
3.3
±0.8
±0.4
±0.6
2.2
5.4
3.1
0.0024
0.0021
0.0060
0.37
0.41
0.88
45
Web Appendix 2. South patch particle summary, continued
234
Lat.dec
(S)
Long.dec
(W)
Date
(2002)
234
Th
Th
POC
POC
C/N
POC/Th
POC/Th
bSiO2
bSi/Th
Time Depth >54 μm
1-54 μm
>54 μm
1-54 μm molar >54 μm
1-54 μm >54 μm >54 μm
(GMT) (m) (dpm L-1) ±error (dpm L-1) ±error (μmol L-1) ±error (μmol L-1) >54 μm (μmol dpm-1) ±error (μmol dpm-1) (μmol L-1) (μmol dpm-1)
0-50m - OUT
66.61
171.82
66.59
172.23
66.32
171.39
66.03
170.99
66.17
172.63
66.58
172.99
65.25
172.72
65.90
170.74
65.87
170.60
31Jan
02Feb
06Feb
13Feb
17Feb
18Feb
18Feb
19Feb
20Feb
0820
0311
0030
0105
2315
0455
2330
2347
0240
20
20
19
19
17
14
16
16
47
0.093
0.144
0.015
0.031
0.151
0.063
0.010
0.098
0.167
±0.001
±0.002
±0.000
±0.001
±0.001
±0.001
±0.000
±0.001
±0.001
0.527
0.411
0.526
0.569
0.277
0.122
0.098
0.221
0.444
±0.005
±0.004
±0.004
±0.005
±0.005
±0.002
±0.002
±0.004
±0.006
0.446
0.745
0.051
0.152
0.973
0.494
0.151
0.517
0.969
±0.012
±0.089
±0.008
±0.022
±0.041
±0.102
±0.023
±0.018
±0.039
2.3
2.4
2.4
2.3
2.0
1.5
1.7
1.6
2.9
7.1
6.3
4.3
7.9
5.6
3.7
6.0
6.3
6.3
4.8
5.2
3.4
4.9
6.4
7.9
15.4
5.3
5.8
±0.1
±0.6
±0.5
±0.7
±0.3
±1.6
±2.4
±0.2
±0.2
4.4
5.9
4.5
4.1
7.1
12.5
17.1
7.3
6.6
0.0595
0.0711
0.0067
0.0253
0.0812
0.0687
0.0156
0.0372
0.1144
0.64
0.49
0.44
0.81
0.54
1.10
1.59
0.38
0.69
50-100m - OUT
66.61
171.82
66.59
172.23
66.32
171.39
66.03
170.99
66.17
172.63
66.58
172.99
65.25
172.72
65.90
170.74
31Jan
02Feb
06Feb
13Feb
17Feb
18Feb
18Feb
19Feb
0820
0311
0030
0105
2315
0455
2330
2357
68
53
82
77
72
69
71
71
0.023
0.186
0.002
0.012
0.013
0.044
0.007
0.093
±0.001
±0.002
±0.000
±0.001
±0.000
±0.000
±0.000
±0.001
0.291
0.490
0.153
0.206
0.313
0.330
0.263
0.372
±0.004
±0.005
±0.003
±0.003
±0.005
±0.005
±0.003
±0.005
0.131
0.972
bd
0.065
0.131
0.141
0.052
0.423
±0.006
±0.038
5.2
7.6
bd
6.4
6.7
7.1
6.9
6.7
5.6
5.2
±0.3
±0.2
±0.010
±0.004
±0.023
±0.012
±0.090
1.0
2.5
0.7
0.9
1.8
1.2
1.1
1.8
5.4
9.9
3.2
7.4
4.5
±0.9
±0.4
±0.5
±1.7
±1.0
3.5
5.2
4.3
4.4
5.8
3.7
4.3
4.9
0.0752
0.0647
0.0008
0.0036
0.0034
0.0119
0.0053
0.0460
3.20
0.35
0.49
0.29
0.25
0.27
0.75
0.49
>100m - OUT
65.87
170.60
20Feb
0240
122
0.003 ±0.000 0.153 ±0.003 0.021 ±0.010
0.5
bd
6.2
±2.9
3.4
0.0009
0.27
a = POC blank >75%
bd = below detection
46
Table 2. Particle flux results
IN
Depth
(m)
Starta±error Day 10±error Day 20±error
234
Th flux (dpm m-2d-1)
25
249±44
563±49
638±68
50
453±90
1117±90
1270±119
75
477±141
1725±140
2049±171
100
350±196
1926±198
2335±234
125
271±250
1999±249
2435±261
POC flux (mmol m-2d-1)
25
2.3±0.7
50
3.0±0.9
75
2.1±0.8
100
1.6±0.9
125
1.2±1.1
bSi flux (mmol m-2d-1)
25
0.2±0.1
50
0.3±0.2
75
0.2±0.1
100
0.2±0.1
125
0.1±0.1
5.3±1.3
7.5±1.8
7.6±1.8
8.5±2.1
8.9±2.3
0.5±0.2
0.8±0.4
0.8±0.4
0.9±0.4
1.0±0.4
All times in days post iron addition
a = Start: day 5.8 IN; day 6.8 OUT
b = End: day 27.4 IN; day 26.6 OUT
c = Units: μmol dpm-1
6.0±1.5
8.5±2.1
9.1±2.2
10.3±2.5
10.8±2.7
0.5±0.3
0.9±0.4
1.0±0.4
1.1±0.5
1.2±0.5
OUT
Endb±error
697±84
1392±137
2309±189
2662±259
2783±263
6.6±1.7
9.3±2.3
10.2±2.7
11.8±3.2
12.3±3.3
POC/Thc±error
9.4±2.2
6.7±1.5
4.4±1.0
4.4±1.0
4.4±1.0
0.6±0.3
0.9±0.5
1.1±0.5
1.3±0.6
1.3±0.6
bSi/Thc±error
0.81±0.40
0.67±0.34
0.48±0.20
0.48±0.20
0.48±0.20
Depth
(m)
Starta±error
Endb±error
25
50
75
100
125
291±41
630±82
694±131
528±188
382±246
309±41
639±82
826±131
778±188
721±246
25
50
75
100
125
25
50
75
100
125
1.9±1.0
4.0±2.2
4.2±1.8
3.2±1.7
2.3±1.7
0.2±0.1
0.4±0.2
0.3±0.2
0.3±0.1
0.2±0.1
2.0±1.1
4.1±2.2
5.0±2.1
4.7±2.2
4.3±2.3
POC/Thc±error
6.4±3.4
6.4±3.4
6.0±2.4
6.0±2.4
6.0±2.4
0.2±0.1
0.4±0.2
0.4±0.2
0.4±0.2
0.3±0.2
bSi/Thc±error
0.81±0.40
0.67±0.34
0.48±0.20
0.48±0.20
0.48±0.20
47
Table 3. Comparison between SOFeX and AESOPS: mixed layer properties and 100m export fluxes
Units
Date
Latitude
Longitude
Temperature
Salinity
Mixed layer depth
N-inorganic:
(NO3 and NO2 and NH4)
Silicate (dissolved)
Chlorophyll a
Biogenic Si (particulate)
POC
Primary production
Total thorium-234
POC/234Th at 100m
234
Th flux at 100m
POC flux at 100m
bSi flux at 100m
o
S
W
o
C
o
m
mol L-1
mol L-1
mg m-3
mol L-1
mol L-1
mmol m-2 d-1
dpm L-1
mol dpm-1
dpm m-2 d-1
mmol m-2 d-1
mmol m-2 d-1
SOFeX
AESOPS
SOFeX
AESOPS
SOFeX
start-IN a KIWI 8 sta 3 end-OUT a KIWI 9 sta 9 end-IN a
30Jan02
16Jan98 19Feb02
28Feb98 20Feb02
66.34
67.78
65.91
66.10
65.85
171.96
170.11
170.79
168.67
171.95
-0.55
0.44
-0.26
0.99
-0.29
33.79
33.56
33.88
33.84
33.91
35.0
19.8
55.0
57.4
55.0
26.8
64.8
0.63 b
1.4
7.8
43 b
2.02
4.4
350
1.6
0.2
25.5
60.1
0.90 c
3.8
16.1
80 c
1.14
5.2
3066 d
16.0
nd
30.2
55.5
0.34 e
0.6
15.2
19 e
1.92
6.0
778
4.7
0.4
23.7
52.8
0.18 f
8.2
13.8
17 f
0.85
5.2
4030 g
20.8
6.9
26.4
50.4
3.8 h
1.0
nd
82 h
1.86
4.4
2662
11.8
1.3
a. Start and end SOFeX dates correspond to first and last thorium-234 sampling dates for all parameters, unless otherwise noted
b. 29Jan02, SOFeX M04, CTD13 integrated to 50m
c. AESOPS Primary Productivity and Chl a estimated from SeaWIFS Chl & model (Buesseler et al. 2003)
d. SS model sta 3, 17Jan98 (Buesseler et al. 2001b)
e. 12Feb02 SOFeX M36, TM46 integrated to 50m
f. Table 1 of Hiscock et al. (2003) mean of 3 subpolar regime stations
g. NSS model 65.5-68 S 20Jan - 01Mar98 average (Buesseler et al. 2001b)
h. 14Feb02 SOFeX M43, TM52 integrated to 50 m
48
FIGURE LEGENDS
Figure 1. A time-series of six surface transects of total 234Th activity, chlorophyll a
concentration and SF6 concentration (squares and shading) collected during SOFeX. The time
over which sampling took place is noted in the upper left of each panel. Dashed horizontal line
represents 238U activity and 234Th < 238U indicates export on sinking particles. Samples were
collected either from 25 m or surface, and on one occasion, selected stations had 234Th data from
both depths for comparison (day 22.9-23.2). Note SF6 vertical scale decreases with time, due to
natural dilution and air-sea losses of this tracer. Chlorophyll a scale also changes to show
progression of biomass response relative to SF6. Transects are plotted either vs. latitude or
longitude depending upon predominant orientation of the transect (see Figure 2 map).
Figure 2. Map of SOFeX site, showing relative locations of 6 transects sampled in Figure 1.
These transects were collected over a 20 day period and represent different stages of the evolving
and drifting patch (Figure 1). For reference, the clearest satellite ocean color image for 16
February 2002 is overlain on these transects to show the relative scale of the patch, but note that
both the location of the patch and its aerial extent would have changed throughout the
experiment.
Figure 3. Depth profiles from (A) the first IN station (t=5.8 days), (B) the last IN station
(t=27.4), and (C) the last OUT station (t=26.6 days). Plotted vs. depth are CTD data for
temperature and salinity. Total 234Th activity and chlorophyll a concentration data are from
same CTD cast. The last IN and last OUT SF6 samples (squares and shading) are also from these
same casts, while SF6 from t=5.8 days are from a collocated station. All horizontal scales are
49
constant, except for SF6 which is decreased for the last two stations due to lower expected SF6
late in the experiment. Details of ship and station ID can be found in Web Appendix 1 at
http://www.aslo.org/lo/toc/vol_xx/issue_x/xxxxa1.pdf.
Figure 4. Total 234Th activities plotted vs. density for the last IN station and last OUT station.
Thorium-234 activities are plotted with associated error.
Figure 5. Complete SOFeX total 234Th time series for IN and OUT data. Vertical dashed line is
238
U calculated from measured salinity.
Figure 6. Time series of the integrated total 234Th activity from progressively deeper layers for
SOFeX. All data with >3 vertical sampling points were plotted and depth averaged using
trapezoid integration (data in Web Appendix 1 at
http://www.aslo.org/lo/toc/vol_xx/issue_x/xxxxa1.pdf). Solid line is the best fit to a linear
decrease in total 234Th vs. time IN and dashed lines represent 95% confidence limits around this
slope (R2, slopes and errors provided in Table 1). Errors on the 234Th activity measurement are
shown for the first and last IN point for each depth interval.
Figure 7. Plots of particle flux vs. depth for (A) 234Th IN, (B) 234Th OUT, (C) POC IN, (D)
POC OUT, (E) bSiO2 IN, and (F) bSiO2 OUT in units of dpm m-2 d-1 for 234Th and mmol m-2 d-1
for POC and bSiO2. Plotted are the estimates for each 25 m layer at the “start” of the experiment
(day 5-6; open circle) and end (day 27; filled circle) of SOFeX, and errors on this flux estimate
50
are shown by the horizontal error bars. For 234Th IN, estimates of the flux on days 10 and 20 are
also plotted. Complete flux data can be found in Table 2.
Figure 8. Time series progression of POC/234Th on large (>54 m) and small particles (1-54
m). These are separated in by depth (0-50 m, panels A and B; 50-100 m, panels C and D) and
by IN (panels A and C) vs. OUT (panels B and D) stations. The average value for the >54 m
particles for the entire time sequence is shown by a horizontal dashed line.
Figure 9. Particulate ratios of (A) POC/234Th and (B) bSiO2/234Th vs. depth for large >54 m
particles both in units of mol dpm-1. IN stations are shown by open circles. The solid line
depicts the best fit used to calculate fluxes. OUT station data are shown only for comparison.
Figure 10. The export efficiency of POC versus biogenic silica at IN (filled circles) and OUT
stations (open squares). Rates of primary production and silica production were integrated to 50
m and weekly averages computed for each of the first three weeks following the initial Fe
fertilization (data from R. Barber, V. Lance and M. Brzezinski). Each weekly averaged
integrated production rate was then compared to the average rate of export at 50 m for the same
time period (Table 2). 50 m was chosen as the depth best representative of the mixed layer, the
approximate extent of Fe-enrichment in the IN water column, which was about the 10% light
depth. Lines drawn represent ratios of carbon:silica export efficiencies of 1:1 (solid line) and
1.5:1 (dashed line). Errors were determined from the standard error on the weekly export
efficiency ratio (data are averages from n= 40, 11 and 10 stations for primary production, bSiO2
uptake and export, respectively).
51